|
{ |
|
"best_metric": 0.7673333333333333, |
|
"best_model_checkpoint": "Bert_tinybert-distilled/run-10/checkpoint-344", |
|
"epoch": 2.0, |
|
"eval_steps": 500, |
|
"global_step": 344, |
|
"is_hyper_param_search": true, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 1.0, |
|
"grad_norm": 1.0083345174789429, |
|
"learning_rate": 0.0006048203244311811, |
|
"loss": 0.5344, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"eval_AUC": 0.8101155555555555, |
|
"eval_BAC": 0.744, |
|
"eval_MCC": 0.5208119231436024, |
|
"eval_PR_AUC": 0.7682563940585828, |
|
"eval_accuracy": 0.744, |
|
"eval_loss": 0.4804428815841675, |
|
"eval_runtime": 6.2325, |
|
"eval_samples_per_second": 240.676, |
|
"eval_steps_per_second": 7.541, |
|
"step": 172 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"grad_norm": 1.5728729963302612, |
|
"learning_rate": 0.0005040169370259843, |
|
"loss": 0.509, |
|
"step": 344 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"eval_AUC": 0.8224568888888889, |
|
"eval_BAC": 0.7673333333333333, |
|
"eval_MCC": 0.545844031545267, |
|
"eval_PR_AUC": 0.7802114167192694, |
|
"eval_accuracy": 0.7673333333333333, |
|
"eval_loss": 0.4665648639202118, |
|
"eval_runtime": 6.2386, |
|
"eval_samples_per_second": 240.44, |
|
"eval_steps_per_second": 7.534, |
|
"step": 344 |
|
} |
|
], |
|
"logging_steps": 500, |
|
"max_steps": 1204, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 7, |
|
"save_steps": 500, |
|
"stateful_callbacks": { |
|
"TrainerControl": { |
|
"args": { |
|
"should_epoch_stop": false, |
|
"should_evaluate": false, |
|
"should_log": false, |
|
"should_save": true, |
|
"should_training_stop": false |
|
}, |
|
"attributes": {} |
|
} |
|
}, |
|
"total_flos": 900755460000.0, |
|
"train_batch_size": 32, |
|
"trial_name": null, |
|
"trial_params": { |
|
"alpha": 0.8204572075273191, |
|
"learning_rate": 0.000705623711836378, |
|
"num_train_epochs": 7, |
|
"temperature": 7 |
|
} |
|
} |
|
|